LLM EXPLORER 55,709 MODELS INDEXED

DeepSeek V3.1 AWQ W4AFP8 by TMElyralab

By TMElyralab · 6 downloads
HF Repo ↗ Copy URL

DeepSeek V3.1 AWQ W4AFP8 is an open-source language model by TMElyralab. Features: 676.3b LLM, VRAM: 225GB, Context: 160K, License: mit, Quantized, LLM Explorer Score: 0.2.

  Arxiv:2412.19437   4-bit   Awq   Conversational   Custom code   Deepseek v3   Endpoints compatible   Quantized   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/TMElyralab/DeepSeek-V3.1-AWQ-W4AFP8 

DeepSeek V3.1 AWQ W4AFP8 Parameters and Internals

LLM NameDeepSeek V3.1 AWQ W4AFP8
Repository 🤗https://huggingface.co/TMElyralab/DeepSeek-V3.1-AWQ-W4AFP8 
Model Size676.3b
Required VRAM225 GB
Updated2026-07-17
MaintainerTMElyralab
Model Typedeepseek_v3
Model Files  5.0 GB: 1-of-73   5.0 GB: 2-of-73   5.0 GB: 3-of-73   5.0 GB: 4-of-73   5.0 GB: 5-of-73   5.0 GB: 6-of-73   5.0 GB: 7-of-73   5.0 GB: 8-of-73   5.0 GB: 9-of-73   5.0 GB: 10-of-73   5.0 GB: 11-of-73   5.0 GB: 12-of-73   5.0 GB: 13-of-73   5.0 GB: 14-of-73   5.0 GB: 15-of-73   5.0 GB: 16-of-73   5.0 GB: 17-of-73   5.0 GB: 18-of-73   5.0 GB: 19-of-73   5.0 GB: 20-of-73   5.0 GB: 21-of-73   5.0 GB: 22-of-73   5.0 GB: 23-of-73   5.0 GB: 24-of-73   5.0 GB: 25-of-73   5.0 GB: 26-of-73   5.0 GB: 27-of-73   5.0 GB: 28-of-73   5.0 GB: 29-of-73   5.0 GB: 30-of-73   5.0 GB: 31-of-73   5.0 GB: 32-of-73   5.0 GB: 33-of-73   5.0 GB: 34-of-73   5.0 GB: 35-of-73   5.0 GB: 36-of-73   5.0 GB: 37-of-73   5.0 GB: 38-of-73   5.0 GB: 39-of-73   5.0 GB: 40-of-73   5.0 GB: 41-of-73   5.0 GB: 42-of-73   5.0 GB: 43-of-73   5.0 GB: 44-of-73   5.0 GB: 45-of-73
AWQ QuantizationYes
Quantization Typeawq
Model ArchitectureDeepseekV3ForCausalLM
Licensemit
Context Length163840
Model Max Length163840
Transformers Version4.47.0
Vocabulary Size129280
Torch Data Typebfloat16